Vibrotactile signals have the potential to communicate a wide range of information, thus offloading other currently overloaded channels like vision and audition and enhancing the effectiveness and versatility of multimodal interfaces. Most recently, tactons (tactile icons), i.e., abstract structured vibrotactile messages that are analogous to visual icons and auditory earcons, have received considerable attention. To ensure the intuitiveness and robustness of tactons and to reduce the training needs associated with these signals, it is critical to examine how people perceive and interpret tactile signals, to what extent users agree on their interpretation, and to what extent the perceived perceptual and semantic properties of a signal correlate with each other. Very few studies to date have attempted to address these questions. To help fill this gap, a total of 19 participants (UM students) completed a 30-minute session during which they were presented once each with 16 vibrotactile signals in randomized order. These signals were presented via C2 “tactors� (manufactured by Engineering Acoustics, Inc.) and varied along the dimensions of frequency ( between 30Hz-350Hz), amplitude (2-4), gain (1-4; with a gain of 4 translating to an approximately 1mm displacement of the skin), and waveform (sinusoidal, heterodyned with low frequency signals or with similar frequency signals, square/pulsing). A reference stimulus with a frequency of 250Hz, a gain of 3 and a duration of 1000ms preceded each experimental stimulus. The participants were asked to describe the tactile signals in terms of 1) the sensation they triggered, and 2) their assumed meaning. They were asked to provide at least one adjective for each of these two dimensions. The analysis of the resulting data indicates that the adjectives used to describe the sensation relate to the following six dimensions: strength, time, texture, salience, dynamics and annoyance. Each dimension, in turn, consists of several sub-dimensions (e.g., sharpness, texture, softness, and roughness for the dimension of texture). The adjectives that were used to describe the assumed meaning of each signal fall into six categories as well: excitement, urgency, alarm, information, pleasantness and strength. Again, most of these categories include several sub-dimensions, such as importance, urgency, and danger for the category of urgency. The six pulsing and heterodyned signals of similar frequencies were perceived by 27-43% of the participants as relating to the temporal dimension of ‘dynamics’. Between 14-29% of subjects interpreted them as urgent or alarming. The level of perceived urgency increased with the difference between the two frequencies of the heterodyned signal. High frequency and high gain signals (350 Hz or a gain of 4) were most often perceived as relating to strength and urgency. Signals with a high frequency were also associated with the emotive quality of ‘unpleasant’. As many as 57.6% of the participants described stimuli with both a frequency of 350 Hz and a gain of 4 as urgent; 35.7% of the subjects interpreted them as alarming, and 21.4% referred it to them as being uncomfortable. In contrast, low frequency tactons (below 150 Hz) tended to be interpreted as weak, soft, and slow, as well as pleasant or comfortable. Finally, more than 40% of the participants describe the two signals with a gain lower than the reference signal as soft, weak or unnoticeable. Overall, the findings from this study reveal a number of rather robust natural mappings between properties of tactile stimuli and their perceived meaning. They help designers of tactile and multimodal interfaces by taking a first step towards creating a tactile vocabulary and by identifying stimuli that should be avoided because of large interindividual differences in interpretation.